Dynamics Model Confidence Testing

Dynamics model confidence testing quantifies the credibility of the dynamics model by comparing real vehicle data against simulation results. The typical workflow is: collect real vehicle measurement data under various operating conditions, perform parameter calibration (adjust dynamics parameters) on the dynamics model, then use the calibrated model to simulate the same conditions, compare simulation data with measured data, and quantify dynamics confidence.

Taking a gasoline-powered vehicle as an example, the commonly used operating condition ranges are as follows:

• Powertrain performance tests (5 categories, 21 sub-conditions): Full-throttle acceleration, partial-throttle acceleration, constant-speed straight-line driving, variable-throttle acceleration (multiple throttle openings × multiple pedal application rates), and idle coasting.
• Braking performance tests (4 sub-conditions): Deceleration at 100 km/h with 10%/30%/50%/100% brake pedal application.
• Handling and stability tests (5 sub-conditions): Steady-state circular driving, step steering, ramp steering, sinusoidal steering, and static steering.
• Key observations: Vehicle speed, longitudinal/lateral acceleration, yaw rate, wheel speed, engine torque/RPM, steering torque, etc.